Casing programs of many wells in the deepwater (>1,000 ft of water) Gulf of Mexico, the Nile delta, and other offshore drilling provinces are affected by shallow water flow geopressure (SWFG) within 1,000-2,500 ft of the seafloor. When SWFG sands are not properly isolated, a sustained brine flow may occur around the surface casing, leading to buckling of the casing and a large deposit of sand carried to the seafloor by the flow.
Over the past 20 years, oil industry operators, contractors, and regulators have made a concerted effort to reduce subsea wellbore and facility losses caused by near (1,000-2,500 ft below mudline) seafloor uncontrolled brine flows from unconsolidated sands. These SWFG are primarily associated with wells drilled in greater than 1,000 ft of water. SWF events ultimately lead to buckling of casing strings, as structural support of surrounding sediments is lost.
Pre-drill evaluation of SWFG-prone sands involves regional setting and drill location choice. Regional context for the probability and severity of SWFG events can be developed and mapped from previous drilling experience and sedimentation rate. Appropriate choice of a drilling location involves both the recognition of sands from seismic character and mapping of the sealing and/or trapping geometries that can contain overpressures. Regional context and drill location evaluation are inferential, reducing but not eliminating SWFG risks. Progress toward quantitative prediction of shallow geopressures from surface seismic continues to present, but this approach is challenged by the need for innovative acquisition and calibration methods.
Well design and drilling practices include casing depths that top-set and/or immediately case-off SWFG sands, monitoring of well behavior to allow rapid kill, drilling with calcium chloride brines or other weighted fluids, and cementing practices.
Ozona SWF event
The Ozona prospect (Garden Banks block 515) is in 3,300 ft of water in the deepwater Gulf of Mexico. The prospect is in a greater Auger salt rimmed mini-basin between two producing fields operated by Shell (SEPCO). Ozona is 6 mi south of the Auger field (Garden Banks block 426) and 6 mi northwest of Shell’s Macaroni field (Garden Banks block 602). The offset fields place the Ozona SWF sands in a regional context.
A shallow (10,000 ft) exploratory well at the Ozona prospect encountered and controlled the SWFG sands of the Sangamonium age in June 1999. After drilling the SWF interval, casing was run to 5,000 ft MD to put the shallow sands behind pipe. More than two years later a second exploratory well (Garden Banks block 515 No. 2), designed to reach objectives deeper than 20,000 ft, was spudded 6,000 ft to the southwest of the No. 1 well. Although the shallow, geopressured sands were controlled during drilling operations, the No. 2 well was lost to a subsequent SWFG event after the SWFG sands were cased off.
For future avoidance of this costly and challenging drilling hazard, this SWFG event was evaluated through five data sources:
• Near offset wells with MWD logs
• Cased hole formation pressure measurements
• Laboratory measurements on the expelled sand
• Conventional 3D seismic acquired before the event
• A second 3D seismic survey acquired after the event.
By reconciling the available data into a quantitative evaluation of the SWFG event in the second Ozona well, this analysis should be of interest to drilling engineers and to geoscientists providing risk assessments for deepwater well locations. The analysis focuses on the characterization of the SWFG sand rather than on specific laboratory, seismic, or engineering methods. There may be significant variability between SWFG events, but the analysis suggests the problem of shallow hazards will remain challenging. In the Ozona SWFG event, sands with minimal geopressures and limited aerial extent presented drilling problems that had to be addressed through well planning.
MWD correlations
Using MWD gamma ray and resistivity logs, the SWFG sands can be correlated to the north from Ozona into the Auger field and to the south into the Macaroni field. The SWFG sands were subdivided into four parts, labeled S1-S4 (Sands 1 through 4). A red line at the top of each unit and a black line at the base of the sand unit correlate these sand units between Augur, Ozona, and Macaroni. Two of the sand units (S1 and S4) appear to thin and develop a silty log character in the Macaroni log at the southern end of the well log correlation section. The total sand package actually straddles the Sangamonium maximum flooding surface, which is near the base of S1. By depth of burial, the sedimentation rate exceeds an average of 10 ft/1,000 yr, meeting the criteria of rapid sedimentation rate. If renewed sedimentation is correlated to sea level lowering after the Sangamonium transgressive event, calculated sedimentation rates would be even greater.
Details of the SWFG sands are apparent on sections from the 3D seismic data. Each of the sands (S1-S4) is correlated to a blue peak on the seismic section (Runsum display, with the top of the sand approximately at the top of blue, higher impedance peak). The No. 2 well finds the total sand section thicker, or expanded, relative to the No. 1 well. S1 is more reflective, and channelform cutouts that isolate portions of the sands encountered in the wells laterally limit both S1 and S2. East of the No. 1 well, active faults (over salt bodies rimming the mini-basin) create an upthrown, trapping geometry for the Sangamonium sands.
Pressure measurements
Cased hole pressure measurements, acquired in the Ozona No. 1 well, establish the magnitude of the geopressure. The measurement points were recorded on the well log correlation section. The overpressure is determined by comparing the measured pressure to a calculated pressure based on the pressure of a column of seawater to the seafloor plus a column of formation brine (taken as 50,000 ppm NaCl, based on resistivity measurements and porosity data described later in this study). The overpressure calculations suggest S1 and S2 are isolated from each other and the package consisting of S3 and S4. For S1, the thin shale at the base, which correlates to the Sangamonium marine transgression, holds more than 20-psi pressure differential.
For well planning purposes, the cased-hole pressure measurements are repositioned to the formation tops (the top of each sand unit, S1-S4, in this case, as the internal pressure barriers have been established). The results indicate that for riserless drilling, a kill mud of greater than 11 ppg would be required for well control.
Laboratory measurements
After the Garden Banks block 515 No. 2 well was lost to the SWFG event, an ROV monitored the drill site until flow ceased about nine months later. As part of the monitoring program, a sand sample was collected from the cone of expelled sand around the location. Analysis showed poorly sorted, fine grain sand composed of 70% quartz, 20% plagioclase feldspar, and 10% orthoclase feldspar. At the Stanford University Rock Physics Laboratory Dr. M. Zimmer made low stress measurements of porosity, compressional wave velocity, and shear wave velocity. Rather than a single set of measurements at the insitu stress, which is approximately 360 psi, measurements were made over a series of six loading cycles to ever increasing stress. Over the course of the experiment, measured porosity decreases from 0.41-0.36, a loss of 5 porosity units (p.u.). This porosity change is apparently due only to mechanical rearrangement of the sand grains. However, in any single loading cycle, porosity change associated with 100 psi change in stress, which is the amount of overpressure observed in the Garden Banks block 515 No. 1 cased hole pressure measurements, is quite small.
The laboratory measurements on the sand sample have to be adjusted to field conditions to allow direct comparison to the reflectivity variability observed on the seismic. Numerical fluid substitution was performed on all of the measurements, replacing the laboratory fluid with 50,000 ppm NaCl at 65° F. The bulk modulus of the formation fluid is calculated to be 2.53 Gpa (GigaPascals - pressure unit), the bulk modulus of the sand grains is 40.8 GPa, and the shear modulus of the sand grains is 38.4 GPa, based on the mineralogy noted above. Porosity is crossplotted with acoustic impedance, which can be related to seismic reflectivity, based on the adjusted laboratory data at the effective stress approximating insitu conditions. With this crossplot, reflectivity variability can be related to porosity variation in the SWFG sands encountered at Ozona.
Seismic before and after
Two seismic surveys were used to evaluate the Ozona SWF sands. These surveys were chosen for their relatively close line spacing (30 m or less). SEPCO acquired the first survey, Auger 90, in 1990 to support their development activities. The company acquired the second survey in 2002, after drilling the Garden Banks block 515 No. 2 well. A depth structure map on the top of the SWF sands shows both the Ozona No. 1 and No. 2 are more than 400 ft off the crest of a shallow structural high in the southwest corner of Garden Banks block 515. The Ozona No. 2 well is potentially isolated from the No. 1 well by channel incisions at the S1 and S2 levels, but the deeper sands (S3 and S4) are in communication between the two wells.
Inverting the Auger 90 data for acoustic impedance shows the highest average acoustic impedance areas are adjacent to the channel cuts. At the time of the first survey, Ozona No. 2 sits directly to the north of a lower impedance that is bounded by areas of higher impedance. The consistency of the acoustic impedance response along the channel axis, which appears geological and reasonable, suggests the variability in the baseline survey is not an artifact of the data.
When the Auger 90 data are compared to the second survey, the low impedance area is no longer present. The two surveys were placed into a single grid, referenced to the seafloor, and balanced using a matching filter across the seafloor reflector and overburden section, but not including the SWF sands. The increase in acoustic impedance appears only to the south of Garden Banks block 515 No. 2, even though the sand and pressure variations related to the SWF event would also extend to the north. Using the laboratory measurements on the expelled sand, the acoustic impedance change can be related to a 1 p.u. average porosity loss.
To scale the size of the SWF event, the quantity of expelled brine is a suitable measure. However, there has been no practical way to measure these flows. The minimum quantity (with error bars of at least 30%) of expelled brine is estimated from the porosity change from seismic in S1 and S2. From 220 ft of net sand, average compaction of 1.7 ft and 12.6 MMcf of brine expelled would be expected. More brine may flow from areas that did not undergo detectable porosity loss.
Porosity change
Mass balance reconciliation of the Ozona No. 2 SWF event with the seismic and ROV observations is ongoing. Preliminary calculations show that porosity loss in the S1 and S2 sands are critical to both the magnitude and duration of the SWFG event. The comparison of the two seismic surveys suggests such a porosity change occurs at distances up to 3,000 ft from the Ozona No. 2 well. However, the laboratory data do not show significant porosity variation with effective stress variation of 100 psi.
Other research also casts doubt on the quantity of sand that would be expelled by a SWF sand at a depth greater than 1,000 ft below mudline and overpressures of 120-150 psi. On a seismic section connecting the Ozona No. 1 and No. 2 wells, a shallow slump between 1,500 and 1,600 millisecond is visible. If a slump in S1 and S2 were induced, porosity losses under critical stress (shear failure, in this case associated with large shear strain) could explain both elevated pore pressure allowing sand expulsion and the seismic response (porosity loss).
A hypothetical scenario concerning porosity loss of 5 p.u. due to slumping into the annular volume occurs around the well between the times A and B were developed. A pressure increase related to sudden loss of sand porosity triggers, or possibly reactivates, dip-oriented slumping between times B and C. Porosity loss of 1 p.u. within the slumping sand further elevates pore pressure, and the full SWF event is underway.
When an extensive slump occurs within a sand that exists in pressure isolation, a feedback process develops that explains the duration of SWFG events. Geopressures related to compaction disequilibrum are transient, requiring ongoing porosity loss as fluid escapes over geologic time.
It has been a challenge to explain the persistence of SWFG events, but placing the porosity loss within the sand section rather than the bounding shale section is a possible explanation that bears further examination.
Editor’s Note:This is a summary of the SPE 90980 paper presented at the 2004 SPE-ATCE.
